Recommendation API
Every user should see a different catalog.
Ranked content IDs per user, per surface, per session — rendered by your own frontend.
Ranking inputs
Relevance is a combination, not a rule.
Content affinityBehaviorRecent activityPopularityFreshnessContextBusiness rules
Personalization
Two users. One catalog. Two feeds.
Illustrative data
Person A
usr_821 · financeGrowth Finance0.95
Retention0.88
Pricing0.81
Person B
usr_490 · demand generationAcquisition0.95
Paid Media0.88
Attribution0.81
Contract
IDs and scores. Nothing you didn't ask for.
No embedded widget, no iframe, no styling. Your product decides how a recommendation looks.
GET /v1/recommendations
{
"user_id": "usr_821",
"surface": "home_rail",
"results": [
{ "item_id": "course_204", "score": 0.951 },
{ "item_id": "lesson_18", "score": 0.902 },
{ "item_id": "clip_7734", "score": 0.874 }
]
}Your catalog already has the answer.
Help users find it.
Don't Migrate. Connect.